Cerium is not rare. Neither, in any strict geological sense, are the other sixteen elements filed under the label rare earths. The US Geological Survey notes in its February 2026 commodity summary that rare earths are relatively abundant in the Earth’s crust, and that what is uncommon is finding them in concentrations worth digging up. The scarcity that matters here is industrial rather than geological, and it has an address.

The same magnets and coatings turn up in phone speakers and camera modules, electric vehicle traction motors, offshore wind turbine generators, and fighter jet actuators. In 2025, China mined 270,000 tonnes of rare-earth-oxide equivalent out of a world total of 390,000, according to those USGS figures, close to 69 per cent. The more consequential number sits downstream. The International Energy Agency has put China’s share of rare earth separation and refining at about 91 per cent, with Malaysia a distant second. The IEA’s Global Critical Minerals Outlook 2026 records a modest decline in that concentration as American and Malaysian capacity came online, the one critical mineral where things eased rather than tightened last year.

The bottleneck is chemistry, not geology

Australia mined 29,000 tonnes in 2025, second in the world, and much of that concentrate travels to Malaysia to be separated. Separation is the difficult, unglamorous middle of the chain: seventeen chemically similar elements, teased apart across long cascades of solvent extraction stages, at margins thin enough that Western plants closed rather than competed through the 2000s.

The mining figures understate the position further.

The USGS estimates Burma’s 22,000 tonnes from Chinese import data, which is a polite way of saying that ore leaves Kachin State and arrives at Chinese refineries.

The number everyone quotes, and what it rests on

Anyone who has read about rare earths and defence has met the claim that each F-35 requires roughly 417 kilograms of rare-earth material. Adamas Intelligence, a market research firm covering the sector, traced that number in April 2026 to a single September 2012 Department of Defense study on recycling feasibility, submitted to Congress and never released, surfacing publicly as one sentence in a Congressional Research Service report from December 2013. There is no disclosed calculation behind it. It is a number that stuck.

Adamas ran its own first-principles estimate from disclosed component counts and typical magnet loadings, putting the actual content at 40 to 70 kilograms of finished rare-earth-bearing material, of which 11 to 20 kilograms is elemental, and roughly 23 kilograms of that is samarium-cobalt alloy. This is one firm’s audit rather than a settled figure, and it is worth reading alongside a separate data point: the Government Accountability Office reported in September 2024 that the Pentagon’s own rare-earth consumption is a rounding error against global demand, under one-tenth of one per cent.

The defence exposure was never really about tonnage. It comes down to a handful of grades that nothing else replaces: samarium-cobalt, which holds its magnetisation at temperatures where neodymium magnets fail, and the dysprosium and terbium added to high-coercivity neodymium magnets for the same reason. Those are the materials inside high-temperature actuators, precision motors, and pointing mechanisms, and they are also the materials China controls most tightly.

What a shortage actually looks like

Yttrium is the clearest illustration. Chinese customs data compiled by the Center for Strategic and International Studies show 17 tonnes reaching the United States between April and December 2025, against 333 tonnes in the eight months before controls began. Aerospace manufacturers use yttria-stabilised zirconia as a thermal barrier coating on turbine hardware, applied in films measured in hundreds of microns. The quantities involved are trivial; finding a substitute has not been.

Arnold Magnetic Technologies, a magnet manufacturer with an obvious interest in the answer, wrote in February 2026 that licence applications were sitting for 60 to 120 days without decisions, and that a classification of no rejection often meant an indefinite hold. It is a company’s account of its own experience, but it captures the shape of the problem better than any market-share estimate: not a wall so much as a queue with no visible end.

A year of policy, measured against a year of output

China’s April 2025 controls named samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium. October brought an expansion covering five more elements and a rule reaching foreign-made goods containing Chinese-origin material. In November, following the Trump and Xi meeting, Beijing suspended the October package for one year, to 10 November 2026, while the April controls stayed in force with general licences issued to selected exporters.

Washington’s response has been large and fast. CSIS counts more than $7.3 billion committed across five departments and agencies, including the $400 million equity stake the Pentagon took in MP Materials in July 2025, a ten-year floor price of $110 per kilogram for its neodymium-praseodymium output, and a $150 million loan to expand Mountain Pass. Mountain Pass produced 8,900 tonnes of rare earth compounds and metals in 2025, its strongest run in decades, and still around a third of American consumption. Over 2021 to 2024, China supplied 71 per cent of US imports of rare earth compounds and metals.

What to watch

Three dates matter most. The October suspension lapses on 10 November 2026. Defence contractors are due to be barred from using Chinese, Russian, Iranian, and North Korean rare earths from 1 January 2027, a deadline CSIS suggests may not be achievable on current progress. And new magnet manufacturing capacity was due to start producing through the middle of 2026, the point at which announcements start turning into tonnes.

Monthly yttrium and magnet export data will answer the question before any press release does.